Dependable Collaboration of Intelligent System of Trusted Smart Systems

In the era of interconnected and rapidly evolving technology landscape, System of Systems (SoS) are becoming prevalent in various domains, including transportation, smart cities, smart production, healthcare, energy, and avionics. The SoS refers to a concept where several independent constituent systems (CS) interact with each other and make a coalition to achieve a common mission that cannot be achieved by a single system1. Also, there is an increasing trend to make both the CSs smarter and their orchestrations and collaboration to be AI-driven for productivity and efficiency, hence called intelligent SoS (iSoS). For instance, in a smart city, various intelligent systems work together to enhance urban living through improved efficiency, safety, and sustainability. If these collaborative intelligent systems exploit new opportunities and provide new services, at the same time they might compromise the system dependability. The failure of such systems can be catastrophic, causing harm to human life, infrastructure or the environment. For instance, failures in intelligent transportation systems in smart cities can lead to severe consequences, such as traffic accidents, disruptions in public transit services, and failures in emergency response systems, resulting in both monetary and human loss.

This project aims to ensure dependability which is critical for maintaining safety and security across the entire iSoS. Some of the key research challenges related to safety and security in iSoS are:

Safety: 1) In iSoS, CSs interact and depend on each other. Therefore, a fault in one CS can propagate to other collaborative systems and may lead to severe safety consequences; 2) emergent hazards in iSoS is less understood; 3) the CS may operate in a dynamic environment that poses significant challenges to the safety of intended functionalities due to increased uncertainty, and unpredictability; 3) the intelligence in the orchestration of iSoS makes real-time decisions, which sometime can be unsafe and can lead to severe safety concerns.

Security: 1) Multiple CSs in iSoS exchange a large amount of data with each other, which broadens the attack surface and increases the chances of cyber-attacks. Vulnerabilities in one of CS can propagate to other interconnected systems, leading to cascading effects. Exploiting a single vulnerability in one CS may compromise the security of the entire interconnected system.

The DOSTY project aims to provide models and analysis techniques for the realization of Trusted Intelligent System of Systems (TriSoS). The main expected result will be our novel three-layered dependability assurance framework targeting the CS level, orchestration level (SoS level), and operational environment level to ensure dependability (specifically safety and security). This framework includes a safety controller at the orchestration level (SoS level) to ensure safe decisions. It also encompasses a system dependency modelling and fault propagation analysis in both orchestration and CS levels as well as mechanisms to avoid fault propagation into other CSs. Additionally, the framework includes an anomaly detection and mitigation algorithm to analyse system telemetry data to detect deviations from normal behavior and identify potential malicious events at the CS’s perception level and also at the orchestration level to ensure a robust protection against evolving cyber threats and maintaining the integrity of critical operations.

• DOSTY-logo.pdf